Determination of An Optimum Cut-off Point for % fPSA / tPSA

Background: Total serum prostate specific antigen (tPSA) and free prostate specific antigen (fPSA) are known to be useful in the detection of prostate carcinoma (PCa). It has been reported that %fPSA/tPSA is more accurate when it comes to distinguishing PCa from non-malignant conditions such as BPH. The recommended cut-off value of %fPSA/tPSA in western countries is 20-25%. Through this study, we aim to determine an optimum cut-off value for %fPSA/tPSA in an Indian population. Methods: This study was performed at our institution between September 2015 and August 2016. The study population included 181 patients who had prostate enlargement and who then underwent PSA based prostate cancer screening with tPSA and %fPSA/tPSA and whose diagnosis was later confirmed by histopathology. An ROC curve analysis was performed to determine sensitivity, specificity and other performance characteristics. The optimum cut-off value of %fPSA/tPSA was determined from ROC curve using Youden’s index. Result: Malignant histology was seen in 17 (9.4%) cases. ROC curve analysis of %fPSA/tPSA revealed an AUC value of 0.777. The cutoff value of %fPSA/tPSA having the optimum balance between sensitivity and specificity was found to be 12.07% (Sensitivity: 70.6%, Specificity: 84.8%, Positive predictive value: 0.324, Negative predictive value: 0.965, Positive likelihood ratio: 4.631 and Negative likelihood ratio: 0.347). Conclusion: The cut-off value of %fPSA/tPSA obtained from our study (12.07%), which was conducted on a South Indian population, is different from the cut-off values seen in western countries and in many studies conducted in western populations.

www.pacificejournals.com/apalm eISSN: 2349-6983; pISSN: 2394-6466 receiver operator curve analysis. Values of area under the curve were also calculated used the same. A p-value of less than the significance level alpha=0.05 was considered significant. Optimum cut-off value was calculated from the ROC using Youden's index.

Result
The study population comprised of a total of 181 male patients with a mean age of 65.02 ± 8.62 years. The youngest was 45 years and the oldest 90 years of age. Prostate malignancy was detected in 17 cases (9.39%). Of these, no patient was below the age of 50 years and the mean age of patients diagnosed with prostate malignancy is 69.88 ± 8.61 years.

Discussion
PSA was first detected in serum in 1980 and since then it has become essential to the management of prostate cancer (PCa). [10] PSA in the human body is required for liquefaction of semen and it is secreted into the seminal plasma. [11] The release of significant quantities of PSA into the main bloodstream is rare in a healthy individual and as such happens only when there is destruction of the basement membrane of the prostate epithelium. [12] This occurs not only in PCa but also in benign conditions. [2] Hence, increased serum PSA levels are not prostate specific. [2] However, a strong correlation between serum PSA and prostate cancer has been proven. [13] [14] The advent of immunoassays made it possible to measure fPSA in various forms. This made it possible to calculate %fPSA. %fPSA was first utilised by Stenman et al and Christensson et al. [4][ 15] It has been demonstrated to be more efficient in distinguishing or differentiating patients with benign prostate histology from those with malignant histology than serum tPSA levels alone. [16] It has been proven that patients with increased serum tPSA concentrations have a higher probability for PCa. It has also been proven that these same patients tend to have a lower %fPSA than patients with benign prostatic disease. [8] In our study as well, we found that there was a statistically significant lowering of %fPSA values in cases of malignant disease as compared to benign prostate disease.
It has also been studied and demonstrated that %fPSA or f/tPSA helps improve the discrimination between PCa and benign conditions especially in cases where serum tPSA is between 4 ng/mL and 10 ng/mL. This helps reduce unnecessary prostate biopsies by helping identify the cases where the need for a biopsy is clear and present. [17][18] [19] An optimum cut-off value for %fPSA, as with any screening test, is essential as it can lead to better and more accurate detection of PCa. Many studies have been conducted amongst various populations to determine cut-off values for better distinguishing PCa from benign lesions.
Our study shows that a cut-off value of 12% gives the optimum balance between sensitivity and specificity (Sensitivity: 70.6%, Specificity: 84.8%). This 12% cut-off value gives an excellent negative predictive value of 0.965 (96.5%) which means that a patient with a %fPSA value of more than 12% has a 96.5% probability of not having PCa.
A study conducted by Safarinejad et al concluded that a f/tPSA cut-off value of 0.18 (%fPSA = 18%) is optimum having a sensitivity of 94.5%. [20] Partin et al suggested that a cut-off value of 15% would detect all advanced tumours while avoiding 80% of unnecessary biopsies, especially in men whose serum tPSA value lie in the "gray" zone (4 to 10 ng/mL). [21] Catalona et al suggested that a cut-off value of 24% would help detect 90% of PCa and avoid appox 18% of benign disease in patients with a serum tPSA value of 2.6 to 4 ng/mL. [7] In a later update by the same authors, a variety of cut-off values were examined and they concluded that a cut-off value of 25% was optimal. [22] Chun et al suggested using median age-specific cut-off values for %fPSA which ranged from 25-31%, and below which the risk of prostate cancer was high. [23] Suzuki et al. reported a 26% decrease in the number of unnecessary biopsies and a sensitivity of 90% when a cut-off value of 10% was applied. [24] Prcic et al found the best combination www.pacificejournals.com/apalm eISSN: 2349-6983; pISSN: 2394-6466 of sensitivity and specificity (sensitivity = 72.3% and specificity = 50.4%) was at a cut-off of 16%. [25] Yilmaz et al suggested a cut-off of 10% whereas Pourmand et al arrived at 13%. [26] [27] The study conducted by Dalva et al came to a cut-off value of 15%. [28] A study conducted on an Indian population (sample size -101 patients) by Thakur et al determined a cut-off value of 12%. [29]

Conclusion
In conclusion, the current study shows that the optimum cut-off value for %fPSA/tPSA which gives the best balance between sensitivity and specificity is 12.07%. This value is seen to be different from the cut-off values determined by most other studies, but this may be explained by the differing populations that were the subject of these studies, most of which were conducted in Western countries. The study by Thakur et al which was done on an Indian population yielded a cut-off value almost identical to ours. As it stands, further study with larger sample sizes is warranted for confirmation of our findings in the Indian context.

Abbreviations and Symbols:
PCa -Prostate Cancer tPSA -Total serum prostate specific antigen fPSA -Free serum prostate specific antigen %fPSA/tPSA -Serum free-to-total prostate specific antigen ratio/percentage ROC -Receiver Operating Characteristic AUC -Area Under the Curve PPV -Positive predictive value NPV -Negative Predictive Value PLR -Positive Likelihood Ratio NLR -Negative Likelihood Ratio